AI-balloon pops, Pasteur fights H5N1 in cattle, Tetris dreams, and the origins of "shut up and calculate"
Desiderata #25: links and commentary
Today’s Desiderata is for everyone. It’s been a while since I permanently locked the series for paid subscribers, so newcomers to The Intrinsic Perspective might not know what they’re missing. It’s a regular round-up of links and commentary, as well as an Open Thread and Ask Me Anything in the comments (feel free to make use of this). If you get something out of it, please consider subscribing, since the rest this year will be locked.
1/11. Since the last Desiderata, The Intrinsic Perspective published:
How to teach your two-year-old to read. More importantly, why to do it.
(🔒) On reaching the age of the TV anti-hero. Meditations from my late 30s.
I want to read your writing (call for subscriber submissions). Share your work on The Intrinsic Perspective this summer.
RIP to the man who beat the efficient market hypothesis. Reflections on Jim Simons' passing.
2/11. I will be on WBUR (Boston’s NPR) talking about the increase of AI-generated slop on the internet. If you want to tune in, the On Point episode will be 10-11AM EST on Tuesday.
3/11. Sometimes when a patient suffers from epilepsy, the only thing to do is cut out the localized source of the seizures. Most neuroimaging, stuck outside the skull, is crude. So such little tissue samples are the window onto the actual complexity of the human brain. Google, in collaboration with researchers at Harvard, just released in Science a detailed look into such a tiny human brain sample. Here’s from the summary at CNN:
Ten years ago, Dr. Jeff Lichtman — a professor of molecular and cellular biology at Harvard University — received a small brain sample in his lab.
Although tiny, the 1 cubic millimeter of tissue was big enough to contain 57,000 cells, 230 millimeters of blood vessels and 150 million synapses.
“It was less than a grain of rice, but we began to cut it and look at it, and it was really beautiful,” he said…
Eventually, Lichtman and his team ended up with 1,400 terabytes of data from the sample — roughly the content of over 1 billion books.
To get a sense of the complexity revealed by 1.4 petabytes of electron microscopy data, here’s a single human neuron from their model.
What about mapping an entire human brain? That would be another 1,000 times bigger, Lichtman explained, which means the data would amount to 1 zettabyte. In 2016, that was the size of the entire internet traffic for the year… At the moment, Lichtman said, it would not only be difficult to even store that much data, but there would be no ethically acceptable way of sourcing a pristine, well-preserved human brain.
Well, among the subset of people who do cryopreservation, you just need to find someone who gambles it is better to preserve the actual wiring diagram with the technology we have now vs. the risk of freezing and crystallization from other techniques. Honestly, it’s probably not that hard to find someone who wants their brain digitized rather than frozen.
4/11. There was an interesting essay in Nature on how philosophy was kicked out of modern physics and replaced with the paradigm of “shut up and calculate.” This change was much to the chagrin of Einstein, who never stopped using philosophy as his personal loadstar, referring to himself later in life as a “lone traveler” trying to find ways around the inexplicability of quantum physics.
By the 1950s, the physics community had become broadly indifferent to foundational questions that lay outside the mainstream. Such questions were judged to belong in a philosophy class, and there was no place for philosophy in physics. Mermin’s professors were not, as he had first thought, ‘agents of Copenhagen’. As he later told me, his professors “had no interest in understanding Bohr, and thought that Einstein’s distaste for [quantum mechanics] was just silly”. Instead, they were “just indifferent to philosophy. Full stop. Quantum mechanics worked. Why worry about what it meant?”
5/11. The most impressive AI-generated video I’ve seen, and to this day the only one to actually blow me away, turns out to have had a bunch of editing post-generation in really substantial ways.
6/11. Cows across the nation, but especially in Texas, are sick, producing less milk and seeming lethargic, even coughing (I’ve never heard a cow cough but I imagine it’s pretty distinctive). Right now, despite there being detectable levels of the genetic material of the virus in 1 out of 5 samples of pasteurized milk, the USDA is saying there is no need for a milk recall due to the pasteurization process likely killing the virus. In a way, the normalcy of our civilization is now resting entirely on Louis Pasteur (he seems comfortable supporting the weight).
How worried should we be about the H5N1 virus that has leapt from cows into human dairy workers? The first clearly sourced case was recently reported in The New England Journal of Medicine.
In late March 2024, an adult dairy farm worker had onset of redness and discomfort in the right eye. On presentation that day, subconjunctival hemorrhage and thin, serous drainage were noted in the right eye.
I tried to find a prediction market about the possibility of this becoming a larger issue, but was met with only incredibly-sparsely traded markets of differing rephrased questions. This reinforced my confidence in my criticism that most prediction markets will act unreliably due to it being so easy to “spin one up,” leading to the same shenanigans as with low-cap cryptocurrencies. However, whatever the actual risk of human cases are, it’s likely quite low, because:
Luckily, genetic sequencing of the virus doesn't indicate it has evolved to easily spread among humans.
Essentially, if you get squirted with milk from a sick cow’s udder, you might have a problem, but otherwise this likely won’t affect you. Cats, however, aren’t so lucky.
More than half of cats around the first Texas dairy farm to test positive for bird flu this spring died after drinking raw milk from the infected cows, scientists reported this week, offering a window into a toll the virus has taken during its unprecedented spread through the cattle industry.
Reading up on the spread of the virus had me thinking though: what would the response be to such a domestic event in a post-Covid world? Would we actually have learned any lessons or would it just be a horrifically dragging repeat of the same arguments? Personally, I can’t say I have positive answers to those questions.
7/11. As we all know far too well at this point, animal-to-human transmission events are a historical regularity. In fact, new research grants proof to the idea that leprosy passed, at one point, back and forth between humans and squirrels in Medieval England.
To the team’s surprise, the medieval squirrel strain bore an even closer relation to medieval human strains gathered from the site of the leprosy hospital than to those taken from modern red squirrels. This indicates leprosy circulated between squirrels and humans in England during the Middle Ages, possibly “ping-ponging” between the two…
Perhaps this is because…
Middle Ages England was squirrel crazy: People kept the arboreal rodents as pets and used their furs to line and trim clothing.
8/11. I recently had my own latest book The World Behind the World reviewed by Dr. Joel Frohlich, who does interesting research on fetal brain development and consciousness. The review is on the Institute for Advanced Consciousness Studies’ blog, and I suggest checking it out. While Joel is not as radical as I am in thinking that neuroscience is pre-paradigmatic because we lack a theory of consciousness, he does appear willing to think aspects of it are pre-paradigmatic.
While I’m not convinced that we always need to put subjective experience (the intrinsic perspective) first to understand the brain (the extrinsic perspective), one area where this approach makes perfect sense to me is our quest to understand dreaming. Neuroscientists still have virtually no idea why we dream each night, and Hoel’s “overfitted brain hypothesis” from 2021 is a brilliant use of the phenomenology first approach. In a research paper published in the journal Patterns, Hoel has put forth the most convincing theory of dreaming I’ve ever encountered: dreams and their bizarre phenomenology are necessary for us to learn to generalize from our daily experiences to novel situations. Even if neuroscience isn’t preparadigmatic, the field of dream research certainly was, in my view, until Hoel’s theory.
9/11. Speaking of the Overfitted Brain Hypothesis, Robert Long brought this to my attention: a case of a Redditor who studied Anki flashcards so much he started to dream about them in what’s called the “Tetris effect.”
So my biology finals are in 36 days and about a week ago when school started again I began studying much more intensely than I had during the summer holiday. I pretty much spent every living moment of my life on Anki both making new cards and reviewing old ones…. This amounted to me using Anki for about 14 hours a day on average for the past week. A few nights ago after I was done with studying and went to sleep at about 3:00 am, I hallucinated/saw Anki cards when I closed my eyes and tried to fall asleep. Here's the crazy part: I could see every thought I was having written on an Anki card. Like when I thought to myself "Wow, it's 3:30 and I need to wake up at 6:30 tomorrow", I saw an Anki card with that text written on it. It was the view you see when you're creating a card and you have the two text boxes for the text that appears at the front/back, with half of the text inside the upper text box and half in the lower. Every time I had a new thought I saw an Anki card with that thought written in it, and my thoughts kept changing quickly so the Anki cards kept changing quickly too… I was extremely sleep deprived and I may have been half asleep as I was lying in my bed, trying to fall asleep when it happened, but it was definitely not a dream.
It was a dream. Or a proto-dream, the beginning of a dream, sometimes called “hypnagogic images.” In fact, the “Tetris effect” was discovered by dream researcher Robert Stickgold, who had participants who had never played Tetris before (this was decades ago) play it all day. At night, there was a high prevalence of Tetris-based dreams, where colored blocks would come and go and blend together at sleep onset.
I think dream researchers should pay a lot of attention to such cases, since it is clear that over-learning, like the Redditor’s over-practice of flash cards, is basically the only thing that can reliably trigger dreams with specific content. That’s probably a very good marker of their actual evolved function, which is for preventing overfitting. In no other theory of dreaming (e.g., like that dreams are for emotional regulation or for memory consolidation) is there some clear and reliable connection that matches the content of dreams to the theory (e.g., researchers who think that dreaming is for replaying and consolidating memories have to explain why you usually dream about things that didn’t happen). As I argue for in the paper where I tied together a lot of scattered parts of the literature into the Overfitted Brain Hypothesis (other work on dreaming had dipped its toes but never grounded it in phenomenology, nor even really integrated it in a satisfactory way that I could find in the literature), I think the most important thing is recognizing the point of dreams are the dreams themselves.
10/11. This time last year I was explaining why I think extremely high IQ scores are, well, not actually real. They are simply from either obsessive study of IQ tests (the way contestants study for Jeopardy) or the result of super-variable performances, resulting in scores across poorly-normed tests that don’t correlate well with each other (since most well-studied tests have much lower ceilings).
Your IQ isn't 160. No one's is.
One cannot help but run into people who clearly fantasize about the following scenario: All the great geniuses of the past sit down and take some sort of culture-invariant IQ test, and then we get to line up the numbers and compare them, finally settling once and for all who was the greatest geni…
11/11. As always for the Desiderata series, please treat this as an open thread. Comment and share whatever you’ve found interesting lately or been thinking about below, or ask any questions you might have.
In the 1980's I was a programmer analyst. At times I would spend hours trying to figure out why the logic in my program was not working. Then an odd thing happened, I found at times I would wake up at 2am and the answer would come to me. Then I took it a step further. Instead of banging my head against the wall and staying late at work, I would simply look the program over and go home at 5pm. The next day I would simply have to look at the program and something would tell me that some code or indicator needed to be taken out. The program would then work and I didn't need to bother myself as to logically why. I just went on to my next project. It made my job a lot easier to let my brain figure it out while I slept.
PS. a lot of problems can be solved this way, I call putting it into the brain queue.
Re #9: in high school, I ran track and one day the coach came to me (never sure exactly why) and said we need a pole vaulter for the team. He didn’t have any experience in coaching the pole vault, so he had arranged that I would spend a couple of intensive days learning the basics with another coach at a rival HS who was experienced in teaching the event. We spent about 3 hours or so the first day and then another 3 or 4 the second. Technique-wise, the ‘trick’ to pole vaulting is getting the pole to ‘bend’, which is how you get real height. The whole thing is pretty counterintuitive: you run full speed and jab a gigantic stick into the ground, flip your body upside down while spinning 180 as the world (and your inner ears) tumble around in completely new ways. Well, the night after that second day, after I had first successfully bent the pole, was wild; in my dreams, I was ‘vaulting’ (but not even in a physically realistic way/setting) over and over again as the world contorted in all sorts of unreal ways. It was, phenomenologically, the most unusual dream I can recall. It definitely felt like my brain was trying to figure out how to navigate a new kind of physical space - which it was - very much in keeping with the overfitting hypothesis.